An Intelligent Computer Vision System for Gesture Based Home Automation and Detection of Parkinson's Tremors

For many elderly people with mobility issues, something as simple as turning on the lights becomes a daily struggle. Furthermore, diagnosis of neurological disorders like Parkinson's disease is unaffordable and inaccessible. The main objective of the research was to utilize machine learning to create a home automation system that performs medical monitoring for hand tremors, including Parkinson's tremors. We built a custom dataset with over 700 hand images and adapted the Yolo-v5 model to our application using transfer learning. The resultant model succeeded in classifying hand gestures in real-time, reaching 96% accuracy. We created a prototype of a home automation system, where different gestures were used to control lights. To add medical monitoring capabilities to the home automation system, we proposed a novel hand-tremor detection method. When tested with simulated hand tremors, the model's results matched well with results from an accelerometer, demonstrating its potential for detecting Parkinson's tremors. This research shows that the proposed approach can provide easily accessible home automation and medical monitoring for the elderly. This novel system can help improve the health and safety of the elderly community globally.